Voltage-gated calcium channels (CaV) are critical regulators of excitability, and the L-type channel CaV1.2 is a central determinant of cardiovascular function and an established therapeutic target. CaV1.2 modulators, such as dihydropyridines, target conserved pore regions and lack subtype specificity, limiting their use as therapeutics and as molecular tools to study CaV1.2 activity. We present a deep-learning computational design framework to generate next-generation modulators of CaV1.2 by targeting two distinct structural sites: the extracellular pore and an allosteric region formed by the calciseptine (CaS) peptide toxin binding site. We design protein mini-binders and macrocycles to target these regions, achieving direct pore blockade to inhibit ion conduction and tunable allosteric regulation of channel gating. Our pipelines integrate deep-learning-based generative backbone and sequence design methods (RFdiffusion, ProteinMPNN, and Latent-X), structure prediction methods (AlphaFold2/3 and Boltz-2x), and physics-based approaches (RosettaDock) to evaluate large libraries of designed candidates. Mini-binders and macrocycles are filtered using rigorous in silico criteria, including interface energetics, structural confidence, and docking profiles, to prioritize those with the highest likelihood of experimental success. Promising mini-binders and macrocycles have passed these computational thresholds and are advancing toward experimental validation. This dual approach, targeting both the allosteric regulatory site and the pore with mini-binders and macrocycles provides a platform to dissect CaV1.2 function and broaden strategies for selective therapeutic modulation. More broadly, it demonstrates how combining mini-binder and macrocycle design can expand the druggable landscape of ion channels beyond traditional small molecules.
Gonzalez et al. (Sun,) studied this question.